[time-nuts] My Garmin 18x, Ver 3.50, currently 1 second slow to UTC
David J Taylor
david-taylor at blueyonder.co.uk
Wed Jan 5 10:16:31 UTC 2011
> I think you will have a much better chance of getting Garmin to fix
> if we can come up with a clear description of what is broken and/or a
> recipe for a setup that demonstrates the problem.
> What do you mean by "more than one second delayed"? That sounds like
> you are
> looking at the wrong PPS signal.
> Could you describe your test setup?
> I don't see a problem with a delay as long as it's reasonably stable.
> has fudge time1 to correct for that. (Even the old "working" firmware
> had a
> horrible delay.) Jitter would be a problem.
> What sort of range of delays are you seeing?
> Kiwi Geoff reported about 150 ms of jitter. That's ugly, but I think we
> live with it.
If the delay were always less than one second, and reasonably stable, it
would indeed be something we could live with, and something which could be
compensated for in ntpd, for example.
With earlier versions of the firmware, the serial output started some
hundreds of milliseconds after the leading edge of the PPS signal, and
there was no ambiguity as to which second the serial data referred.
However, with later versions of the firmware that delay has now reached
almost one second, and sometimes is more than one second, leading to an
ambiguity about which to PPS second the serial data refers.
Test setup is an oscilloscope connected to the serial and PPS outputs of
the GPS, with the two channels displays set to "chop" mode for
simultaneous display (i.e. not alternate display). When triggering from
the PPS leading edge, the serial output can be seen drifting across the
PPS signal, with the serial output sometimes starting after the leading
edge. When triggering from the start of the serial output, the PPS signal
starts mostly after the serial output, but sometimes the PPS has already
started when the serial output starts.
So what started as a few hundred milliseconds delay with earlier firmware
has now stretched to a delay of around one second, resulting in an
ambiguity as to which leading PPS edge the serial output refers.
Is that clear enough?
SatSignal software - quality software written to your requirements
Email: david-taylor at blueyonder.co.uk
More information about the time-nuts